PeerDAS in Grandine by Hangleang | Devcon SEA

[Music] okay thank you okay so the motivation behind the project uh because there are uh some interesting uh topic and uh uh chain that uh proposed by the client team um like why I use uh why I pick up p uh to be working on and why is I uh choose a cred in to contribute to so because like uh at the time that I start the program P do uh was considered to be included in the Petra so they are have like a specific deadline and also uh there is uh active research and development and also like a running down that uh uh uh all client team like working on and to implement the the P so yeah but uh due to the like uh the daret instability and the interoperability uh issue that happening in the previous stet that's why move into the faka the network upgrade uh the after the Petra one and also the uh uh um the cring is right and in R because I'm also um familiar with the programming language and I and then you might ask like uh Lighthouse also writing in Dr as well but why I didn't uh contribute uh to Lighthouse anyway but because of the last uh point because tring is a like a new client uh cons client and the team is uh like a small on like only four people specifically so they don't have like uh capacity to handle a different thing yes so that's why I would like to contribute to to them yeah so a little bit of a Contex about the pasas so Pedas is like um the uh networking chain to um the protocol e protocol which allow the datab sampling and appear to do datab sampling without like to download all of the blob that uh in in the network to ensure the blob availability just like uh download just a few Chan so now like without P so H node have to download or and and to participate in the gossip shop the subnet so all the all the block like in in the block that I have like six Block in it so they have to download all of them like that there's no like theability sampling in there so then if like the P activated so like uh the note like only to participate a few CHS of it uh um in in in in the gossip sub so they don't need to download the whole the block data just a few CH like like in the diagram B so the project uh progress so just want to mention like uh before I start working on P like graning team have done the ground work um already already joining the the first interet of the p and uh um because like the spec is uh involving and changing so like uh they like I mentioned earlier like the team like don't have the capacity to handle of different things um and and then I uh join to contribute to and so I make work as uh some uh done some work on and then ready to join the Pedas D to so there is a switching into like switching from the pust um a peer sampling to the subnet sampling so it is kind like a simplified version of the PE P dust so uh like uh PE dust without peer sampling so they don't like uh uh um get it a complexity of the the PSM BL yeah and also another spec chain into the D to as well so there's a add the metadata V3 so it is kind like a OBC method like it is just a simple like add a new field into uh the metadata like the costet count so it is small thing and and fun fact that uh the D two didn't went up where so it um there's a back in gring that uh cause like um then that uh the the whole to be like messed up so I'm the one that messed up and also at least like we found that the H case that uh that that spe is missing like the check in in the the Colum with the zero commitment so uh and then at later on uh we uh relaunched the denet the D Tre so there is a a PO request and the chain to the spe to include the the validation on the uh Z block so yeah at least we found that the the H case on that and then also like the Destry also didn't went uh what as well so there um like a re from on the prism that that um like uh uh re around like 100 uh thousand of slot so there Ste uh some of like uncertainty about the real uh root cause of the issues yet and mainly maybe like around the peers uh scoring yeah and the red limited stuff and after that uh like uh I found some issue regarding to the By Request and response um and also like some of the uh synchronization issue that happening um previously like uh CR like cannot synchronizing from thees yeah and then uh previously uh they don't uh didn't have like that uh reconstruction handling so I uh try to make sure like uh the Reconstruction is have been implementing before the damet 4 and then I um implementing the rebase P onra so it is like um rebase so the team uh there's a spec chain in uh in the D 4 so and another one is the engine G block we want but it's not uh D so I'm working on it uh currently and there are several like open p okay sorry uh there are uh several several open p may be include as well in one of them uh uh the standard L matric uh from one of the fellow here is gaja you working on it oh so the next step I will like continue and contributing to the project until like it with with launch and continue implement the uh engine G blob one it is the uh optional one but I I most of the Cent are already implementing uh this engine apis and can you facing back because there are some few back that Ste um in in the implementation and also continue improving the performance clean up the Cod and yes and there will be uh later later on like d so I will keep up with the spe chain so the challenge is that um it is the the the most challenging and interesting uh part for me to implement uh the pust is mainly like to to understand the concept of datability sampling uh par eally in the pias because like it's not um the M uh the important thing is not uh um the hard part is not uh the implementing like you try to understand need to understand the concept and also like to get yourself uh familiar with the structure of the codebase crin and also all uh all the clients because uh you need to uh understand the behavior the correct Behavior like when you debu you will like find some of like um like uh uh wrong Behavior or something that and then you you you need to um uh check with all the client implementation uh like what is the their behaviors and then uh try to interop yeah and my experience in the EPS yes it it's great for me uh I before I join APS I did not uh understand anything about the call protocol development and I don't know uh about the decision that made into the protocol chain and it's a lot and I I might say I I'm a permissionless participant and and I uh it is a great opportunity like for the public to uh would like to uh uh exploring the core protal development is is completely permissionless yeah yeah thank you thank you everyone all right thank you any questions got one question over here yeah so a couple of questions what do did you use for your testing setup uh for Pierce and another one is uh about what was your uh networking stack oh oh the the for the testing uh we use a CIS for the local testing on so we spin up uh a few note like from a different client so then we we can test the interoperability between clients yeah for the networking state get the same like you SC to see and we add some of the um service to to um doing the mentoring uh monitoring the and a Das and some and aetric something yeah any other questions awesome thanks again Ang [Applause]